Skip to main content

Table 6 Best configuration achieved for a reduction of 75 and 50% memory requirements using McCane and Otte&Nagel metric

From: Multi-GPU based on multicriteria optimization for motion estimation system

Error metric Benchmark Memory MEtime Accuracy
     (Δ ψ)
Barron ‘Translating tree’ 50% 28.6% −0.13
  ‘Diverging tree’ 75% 50.0% 0.00
   50% 33.3% 0.05
McCaneA ‘Translating tree’ 50% 26.9% −0.12
  ‘Diverging tree’ 75% 49.1% 0.00
   50% 28.9% 0.05
McCaneB ‘Translating tree’ 50% 29.6% −0.11
  ‘Diverging tree’ 75% 49.2% 0.00
   50% 34.5% 0.09
Otte&Nagel ‘Translating tree’ 50% 25.3% −0.21
  ‘Diverging tree’ 75% 49.1% 0.00
   50% 35.7% 0.12